Polymicrobial interactions stimulate resistance to host innate immunity through metabolite perception

Abstract

Bacteria in the human oral cavity often grow in an attached multispecies biofilm community. Members of this community display defined interactions that have an impact on the physiology of the individual and the group. Here, we show that during coculture growth with streptococci, the oral pathogen Aggregatibacter actinomycetemcomitans displays enhanced resistance to killing by host innate immunity. The mechanism of resistance involves sensing of the streptococcal metabolite hydrogen peroxide by A. actinomycetemcomitans, which stimulates a genetic program resulting in enhanced expression of the complement resistance protein ApiA. The oxidative stress response regulator OxyR mediates induction of apiA transcription, and this induction is required for coculture resistance to killing by human serum. These findings provide evidence that interaction between community members mediates prokaryotic resistance to host innate immunity and reinforce the need to understand how polymicrobial growth affects interaction with the host immune system.

The Gram-negative bacterium Aggregatibacter actinomycetemcomitans is a common commensal of the human oral cavity and is a causative agent of localized aggressive periodontitis (1). A. actinomycetemcomitans inhabits the mammalian oral cavity beneath the gum line in an area between the tooth surface and the gingival epithelium known as the subgingival crevice (2). A consistent supply of nutrients is provided to the subgingival crevice by a serum exudate referred to as crevicular fluid (3) that passes through the gingiva and flows along the teeth (4–7). Oxygen levels within the subgingival crevice vary greatly, from microaerophilic conditions (2.1 kPa) in the “moderate” pockets (5–6 mm in depth) to near-anaerobic conditions (1.6 kPa) in the “deep” pockets (>6 mm in depth) (8). A. actinomycetemcomitans resides in the moderate pockets of the subgingival crevice and exhibits enhanced growth under microaerophilic conditions (9).

The mammalian oral cavity is home to a robust microbial community composed of many specialized microbes that are well adapted to growth in this environment. As with many complex communities, interactions between individual community members in the oral cavity have a significant impact on phenotypic aspects of the individuals as well as the group (10). Whether the subgingival crevice is healthy or diseased, A. actinomycetemcomitans often resides as a complex surface-associated (biofilm) microbial community, including several species from the genus Streptococcus, such as S. oralis, S. sanguis, S. mitis, and S. gordonii (10–13). These oral streptococci are typically nonpathogenic and rapidly consume sugars within the subgingival crevice, producing the metabolites lactic acid and hydrogen peroxide (H₂O₂). This physiological ability renders oral streptococci extremely competitive in the oral environment because they consume high-energy carbon sources and excrete metabolites that inhibit growth of neighboring microbes (14).

Our laboratory has pursued the idea that because it inhabits environments with oral streptococci (10–13), A. actinomycetemcomitans has adapted survival strategies for exposure to lactic acid and H₂O₂. Indeed, previous studies demonstrated that A. actinomycetemcomitans preferentially utilizes lactic acid over high-energy carbon sources, such as glucose, despite the fact that this bacterium grows significantly more slowly with lactic acid (15). The ability to preferentially use lactic acid not only eliminates caries-causing lactic acid from the oral environment but eliminates the need for A. actinomycetemcomitans to compete with the more numerous and rapidly growing oral streptococci for carbon (10). Instead, A. actinomycetemcomitans has evolved to use the streptococcal metabolic waste product lactic acid for carbon and energy.

Although our previous studies provided insight into the A. actinomycetemcomitans response to lactic acid, essentially nothing is known about how A. actinomycetemcomitans responds to the other primary metabolite of streptococci, H₂O₂. In this study, we examined the A. actinomycetemcomitans response to H₂O₂ by performing a transcriptome analysis of A. actinomycetemcomitans biofilms exposed to H₂O₂. In sharp contrast to other bacterial species, only 2 A. actinomycetemcomitans genes, katA and apiA, were differentially regulated on H₂O₂ exposure. In addition, these genes were regulated in a H₂O₂-dependent fashion during coculture with the oral bacterium S. gordonii. Induction of the outer membrane protein, ApiA, during coculture provided protection of A. actinomycetemcomitans from killing by human serum. Mechanistically, this enhanced protection was enacted by increased binding of the serum protein factor H by ApiA. These results indicate that bacterial resistance to killing by host innate immunity is enhanced during coculture and suggest that A. actinomycetemcomitans utilizes a streptococcal metabolite as a cue to an impending immune response.

Results

A. actinomycetemcomitans Transcriptional Response to H₂O₂.

For gene expression analyses, A. actinomycetemcomitans was grown in a liquid-phase once-flow-through biofilm flow cell (16) and a solid-phase membrane-associated colony biofilm (17). A custom Affymetrix GeneChip microarray (15) was used to monitor gene expression of A. actinomycetemcomitans biofilms in the presence or absence of a sublethal concentration of H₂O₂. Of the approximate 1,800 genes (>90% of the total genes in A. actinomycetemcomitans) that exhibited detectable expression on the GeneChip, only 2 showed statistically significant, reproducible changes in both models on H₂O₂ exposure (Fig. 1A), and these results were verified using RT-PCR (Fig. 1B). This is in stark contrast to similar studies with other bacteria, in which 140–520 genes were differentially expressed on H₂O₂ exposure (18–23).

Fig. 1.

A. actinomycetemcomitans katA and apiA are induced on H₂O₂ exposure. (A) A custom Affymetrix GeneChip was used to examine gene expression of A. actinomycetemcomitans colony and flow cell biofilms in the presence and absence of 1 mM exogenous H₂O₂. Fold changes were determined from 4 pairwise comparisons and determined to be statistically different for katA and apiA (P < 0.05) using GeneChip Operating Software version 1.4. (B) RT-PCR was used to verify katA and apiA induction in colony biofilms on following exposure to H₂O₂. The constitutively expressed gene clpX was used to standardize cDNA template levels, and planktonic-grown bacteria were used to assess the impact of enhanced aeration on basal transcript levels.

The katA gene, which encodes a cytoplasmic catalase (KatA) that directly detoxifies H₂O₂ into O₂ and H₂O, was significantly upregulated in A. actinomycetemcomitans biofilms on exposure to H₂O₂. This was not surprising, because katA homologues in other bacteria are induced on exposure to H₂O₂ (18, 20, 23). More surprising was the H₂O₂-mediated induction of apiA. ApiA is a 33-kDa trimeric outer membrane protein (also referred to as Omp100) (24) that is 55% identical [using pairwise (p) BLAST] to the nonfimbrial adhesin protein YadA in Yersinia pestis and Yersinia enterolitica (25). Similar to YadA function in Yersinia, ApiA stimulates A. actinomycetemcomitans autoaggregation, assists translocation into host cells, and binds the human serum protein factor H (26, 27). Binding factor H inhibits serum complement activity and protects A. actinomycetemcomitans from killing by the alternative complement pathway (26). Although similar in function, yadA and its homologues in other organisms have not been shown to be induced by H₂O₂ at this time.

katA and apiA Are Induced During Coculture with S. gordonii.

Although our results clearly show induction of katA and apiA by direct addition of the streptococcal metabolite H₂O₂, it was not clear if these genes were inducible during coculture with oral streptococci. To examine this, transcriptional fusions of the promoter regions of katA and apiA with the luminescence reporter genes luxCDABE were constructed. These reporter fusions allow the transcriptional activity of katA and apiA to be assessed by monitoring light production. To examine the impact of oral streptococci on katA and apiA transcription, A. actinomycetemcomitans carrying katA-luxCDABE and apiA-luxCDABE was spread on agar plates and light production was examined on exposure to disks containing the oral streptococci S. gordonii. To ensure that any luminescence induction was attributable to S. gordonii H₂O₂ production, 10,000 units of exogenous catalase (or heat-inactivated catalase as a control) was added to each disk. The presence of S. gordonii elicited an increase in both katA and apiA expression by A. actinomycetemcomitans (Fig. 2). This induction was dependent on H₂O₂, because addition of exogenous active catalase mitigated this response (Fig. 2). Induction of each reporter was also observed in planktonic coculture [Fig. S1].

Fig. 2.

katA and apiA are induced during coculture with S. gordonii. A. actinomycetemcomitans containing the katA-luxCDABE or apiA-luxCDABE reporter fusion was spread on agar Petri plates and exposed to S. gordonii–soaked disks containing active catalase (Sg + cat) or heat-inactivated catalase (Sg). Light production was examined using a Syngene imaging system. The image is a composite of visible light and luminescence (blue) photographs. Sterile paper disks did not induce light production by either reporter fusion (not shown).

OxyR Is Required for H₂O₂-Mediated Induction of katA and apiA.

Many H₂O₂-responsive promoters possess operator sequences that bind the LysR-type transcriptional regulator OxyR (28). OxyR exhibits DNA binding activity after oxidation of 2 reactive cysteine residues by H₂O₂ and can act as an inducer or a repressor of transcription. The A. actinomycetemcomitans genome contains an oxyR-like sequence (ORF AA1513) that putatively encodes a protein with 72% identity (assessed by pBLAST) to Escherichia coli OxyR and shows conservation of both reactive cysteine residues (C199 and C208). To determine if the katA and apiA promoter regions contain potential OxyR binding sequences, primer extension was used to map the transcriptional start sites of these genes. The results indicate that the promoters of katA and apiA possess sequences, centered 46.5 and 58 bp upstream of the transcriptional start sites, respectively (Fig. 3 A and B), with significant similarity to the consensus OxyR binding sequence (Fig. 3C) ATAG-n7-CTAT-n7-ATAG-n7-CTAT (29, 30). These sequences suggested that OxyR was a regulator of katA and apiA transcription. To test this, oxyR was insertionally inactivated in A. actinomycetemcomitans and transcript levels of apiA and katA were measured in response to H₂O₂. Inactivation of oxyR abrogated the induction of apiA and katA by H₂O₂, and this activation was restored by expression of oxyR in trans (Fig. 4 A and B, Insets).

Fig. 3.

DNA sequences of the katA (base pairs 154293–154374) (A) and apiA (base pairs 1724157–1724078) (B) promoter regions. Transcription start sites are in boldface, putative −10 regions are underlined, and oxyR-like binding elements are boxed. (C) Alignment of the putative katA and apiA OxyR binding sequences with the consensus binding sequence (29, 30).

Fig. 4.

The A. actinomycetemcomitans oxyR⁻ mutant is hypersusceptible to killing by H₂O₂ and human serum. (A) The H₂O₂ minimum inhibitory concentration (lowest concentration necessary to inhibit visible growth of an organism) of WT A. actinomycetemcomitans (wt), the A. actinomycetemcomitans oxyR⁻ mutant (oxyR⁻), and the genetically complemented A. actinomycetemcomitans oxyR⁻ mutant (oxyR⁻ + oxyR). (Inset) RT-PCR analysis of mRNA levels of katA and the clpX constitutively expressed control in the wt oxyR⁻, and oxyR⁻ + oxyR after H₂O₂ exposure. (B) Survival of wt, oxyR⁻, oxyR⁻ + oxyR, and oxyR⁻ constitutively expressing apiA (oxyR⁻ + apiA) in the presence of 50% (vol/vol) normal human serum. Percent survival was calculated as follows: number of cells recovered from normal human serum treatment/number of cells recovered from heat-inactivated serum treatment. (Inset) RT-PCR analysis of mRNA levels of apiA and the clpX constitutively expressed control in the wt, oxyR⁻, and oxyR⁻ + oxyR after H₂O₂ exposure. As a control, exogenous catalase was added to serum to ensure that any phenotype observed was not attributable to endogenous H₂O₂ within serum. Error bars represent SEM. *P < 0.004 via Student's t test, n = 3.

OxyR Is Critical for Resistance to H₂O₂ and Human Serum.

Based on the observation that OxyR regulates katA and apiA, we hypothesized that inactivation of oxyR would have a profound effect on A. actinomycetemcomitans survival on exposure to H₂O₂ or human serum. To test this hypothesis, we measured the survival of WT A. actinomycetemcomitans and the oxyR⁻ mutant following exposure to H₂O₂ and human serum. Deletion of oxyR significantly decreased survival of A. actinomycetemcomitans on exposure to H₂O₂ and human serum, and expression of oxyR in trans restored survival to WT levels (Fig. 4 A and B). Although it is clear that katA is a critical component of H₂O₂ resistance in numerous bacteria (18–23, 29), it was not known if the decrease in serum resistance was attributable to decreased levels of apiA. To test this possibility, apiA was constitutively expressed in the oxyR⁻ mutant (27). Constitutive expression of apiA restored serum survival of the oxyR⁻ mutant to WT levels (Fig. 4B), suggesting that apiA dysregulation was responsible for enhanced serum killing in this mutant. As expected, expression of apiA had no effect on the sensitivity of the oxyR⁻ mutant to H₂O₂ (data not shown). These results demonstrate that the oxidative stress regulator OxyR is necessary for induction of a gene involved in complement defense in A. actinomycetemcomitans.

Coculture Enhances Resistance of A. actinomycetemcomitans to Alternative Complement Pathway Killing.

Based on the observation that apiA transcription was enhanced during coculture, we hypothesized that coculture with S. gordonii would enhance A. actinomycetemcomitans resistance to serum killing. To test this, A. actinomycetemcomitans serum survival was assessed during coculture with S. gordonii in the presence of heat-inactivated or active catalase. Coculture with S. gordonii enhanced survival of A. actinomycetemcomitans to human serum by ≈7-fold (Fig. 5), clearly demonstrating enhanced survival of A. actinomycetemcomitans during coculture. The presence of active catalase mitigated this increase in survival, implicating S. gordonii H₂O₂ production as the key mediator of this phenotype (Fig. 5).

Fig. 5.

Coculture with S. gordonii enhances A. actinomycetemcomitans resistance to killing by human serum. Fold increase in survival of A. actinomycetemcomitans on exposure to 50% (vol/vol) normal human serum when grown in monoculture (Aa), coculture with S. gordonii + heat-inactivated catalase (Aa+Sg), and coculture with S. gordonii + catalase (Aa+Sg+cat). Ratios were calculated as follows: colony-forming units present after serum treatment/colony-forming units present after heat-inactivated serum treatment. Error bars represent SEM. It is important to note that cell numbers were similar with and without catalase in the heat-inactivated complement cultures. *P < 0.01, **P < 0.05 via Student's t test, n = 3.

Coculture Enhances Factor H Binding to the A. actinomycetemcomitans Cell Surface.

From a mechanistic standpoint, it was not clear how induction of apiA expression during coculture enhanced resistance to serum. Clues were provided by Asakawa et al. (26), who recently demonstrated that ApiA binds the human complement regulatory protein factor H. Factor H is a complement control protein that circulates in human serum, and when bound to cells, it inhibits the alternative pathway of complement activation. Because apiA is induced during coculture, we reasoned that the levels of factor H bound to the A. actinomycetemcomitans outer surface would be increased during coculture with S. gordonii. To examine this, immunofluorescence staining with an anti-factor H antibody was used to quantify the levels of factor H bound to the A. actinomycetemcomitans outer surface during monoculture and coculture growth. Our results revealed that ≈4-fold more factor H was bound to A. actinomycetemcomitans during coculture with S. gordonii as compared with monoculture conditions (Fig. 6). This enhanced binding of factor H during coculture was mitigated by the addition of active catalase, indicating that, as expected, H₂O₂ was the critical cue mediating increased binding.

Fig. 6.

Factor H displays enhanced binding to A. actinomycetemcomitans during coculture with S. gordonii. (A) Immunofluorescent micrographs of factor H attachment to the surface of A. actinomycetemcomitans during monoculture (Aa), coculture with S. gordonii + heat-inactivated catalase (Aa+Sg), and coculture with S. gordonii + catalase (Aa+Sg+cat). Images were recorded at magnification ×1,000. (B) Average green channel fluorescence intensity per cell. Averages were calculated from 40 independent measurements. Error bars represent SEM. *P < 0.0001 via Student's t test.

Discussion

A. actinomycetemcomitans is a common commensal of the mammalian oral cavity, where it resides in a complex microbial community within the subgingival crevice. The subgingival crevice is distinct from the exposed tooth surface and poses several challenges for A. actinomycetemcomitans growth and survival, including competition with faster growing bacteria for nutrients; the presence of antimicrobial serum proteins such as immunoglobins, complement, and antimicrobial peptides (6, 31, 32); and the presence of high levels of metabolites produced by other members of the microbial community. Two of the most prominent microbial metabolites produced in the oral cavity are lactic acid and H₂O₂. As observed with lactic acid (15), our results demonstrate that A. actinomycetemcomitans displays a unique response to H₂O₂ exposure. It is intriguing that only 2 A. actinomycetemcomitans genes exhibited a significant change in gene expression on exposure to H₂O₂ (Fig. 1). This contrasts with transcriptome studies in many other aerobically or microaerophilically grown bacteria, which have shown large numbers of genes (140–520) differentially regulated in response to H₂O₂ (18–23). Many of the H₂O₂-responsive genes identified in these previous studies, such as DNA repair proteins and superoxide dismutases, are expressed by A. actinomycetemcomitans but are not responsive to H₂O₂, suggesting that these genes may be constitutively expressed in A. actinomycetemcomitans as an adaptation to frequent H₂O₂ exposure.

The H₂O₂ response was observed not only on exogenous addition of H₂O₂ but during coculture with the H₂O₂-producing oral bacterium S. gordonii (Fig. 2 and Fig. S1). Not surprisingly, A. actinomycetemcomitans displayed enhanced production of the H₂O₂-consuming enzyme catalase during coculture. This response has been observed previously in several bacterial species and serves to enhance resistance to H₂O₂ (18–23, 29). More intriguing was the finding that A. actinomycetemcomitans also induces the outer membrane protein ApiA during coculture. Because ApiA provides protection from killing by the alternative complement component of innate immunity (26), our data represent a demonstration of a metabolic cue produced by one bacterium mediating enhanced resistance to a component of host innate immunity by another. It should be noted that the mechanism of complement inhibition, namely, enhanced binding of the serum protein factor H (Fig. 6), not only protects the bacterium from the alternative complement pathway but significantly decreases complement-mediated opsonic uptake and killing by host phagocytic cells (33). Thus, it is likely that the impact of coculture on A. actinomycetemcomitans resistance to innate immune effectors extends beyond our observations of protection from the alternative complement pathway.

Based on murine studies, it is likely that A. actinomycetemcomitans is exposed to H₂O₂ produced by streptococci in the oral cavity (14). Why would A. actinomycetemcomitans respond to H₂O₂ by enhancing resistance to complement killing? Although the answer is unknown, in vitro studies reveal that as oral streptococci numbers increase, the levels of streptococcal H₂O₂ increase because of the accumulation of lactic acid (34). This has been shown to have critical consequences in several nonoral in vivo models, because H₂O₂ produced by streptococci (35, 36), in combination with other factors such as lipoteichoic acids (37, 38), induces significant inflammation. Inflammation leads to increased vascular permeability, influx of serum, and recruitment of neutrophils to the subgingival crevice (3). In the context of these previous studies, we propose a model (Fig. 7) in which enhanced levels of streptococcal H₂O₂ produced during early inflammation stimulate a more robust immune response, followed by an influx of innate immune modulators. A. actinomycetemcomitans responds to rising H₂O₂ by enhancing resistance to innate immune effectors. In this sense, A. actinomycetemcomitans utilizes H₂O₂ as an anticipatory signal for an enhanced immune response.

Fig. 7.

Model for the role of H₂O₂ as a mediator of A. actinomycetemcomitans resistance to innate immunity. Enhanced levels of H₂O₂ produced by S. gordonii during plaque growth stimulate inflammation, leading to an influx of innate immune modulators, including complement and neutrophils. A. actinomycetemcomitans responds to rising H₂O₂ by induction of katA and apiA, which, in turn, enhance resistance to innate immune effectors. On recruitment to the site of inflammation, neutrophils increase the levels of H₂O₂ and further stimulate induction of katA and apiA.

Of course, streptococci are not the only source of H₂O₂ in the subgingival crevice. Another source is host tissues, which produce endogenous H₂O₂ from the mitochondria during aerobic respiration (39, 40). Based on the fact that host tissues produce significantly lower levels of H₂O₂ compared with S. gordonii (34, 39, 40) and that most of the host peroxide is likely scavenged by host catalase, we predict that endogenous H₂O₂ production has little impact on A. actinomycetemcomitans gene expression. This is likely not the case for neutrophils, which produce high levels of H₂O₂ on recruitment to the site of inflammation (41, 42). Thus, during gingival inflammation, A. actinomycetemcomitans also faces high levels of neutrophil-produced H₂O₂. Our model (Fig. 7) predicts that before significant neutrophil recruitment to the subgingival crevice, neutrophil H₂O₂ likely has little impact on A. actinomycetemcomitans gene expression. Instead, we predict that during early inflammation, streptococcal H₂O₂ will be the primary stimulant for enhanced production of ApiA and KatA. This initial stimulation by streptococcal H₂O₂ will augment the inflammatory response (35, 36) and stimulate recruitment of neutrophils to the subgingival crevice. We propose that the initial A. actinomycetemcomitans response to streptococcal H₂O₂ not only provides resistance to the influx of alternative complement during inflammation but likely provides resistance to neutrophil-produced H₂O₂. During these latter stages of inflammation, neutrophil H₂O₂ may also serve as an additional stimulus to A. actinomycetemcomitans to enhance expression of katA and apiA. Of course, this model requires vigorous in vivo testing in the future; however, these studies are significantly hampered by the lack of a robust primate model.

Our results demonstrate that A. actinomycetemcomitans uses the S. gordonii metabolite H₂O₂ as a cue to induce expression of katA and apiA, whose products aid in defense against host innate immunity. These findings suggest that A. actinomycetemcomitans may use H₂O₂ as an indicator of an impending host innate immune response in vivo and provide a description of a polymicrobial interaction that influences resistance to host innate immunity. There is considerable interest in understanding how pathogenic microbes evade host innate immunity. In fact, there is substantial effort aimed at developing therapeutics to enhance the effectiveness of the innate immune response (43). Such studies have focused on treatment of monoculture infections, despite the observation that many infections are polymicrobial. Our results clearly show that interaction between 2 prominent oral bacteria significantly affects killing by host innate immunity and reinforce the idea that understanding how polymicrobial interactions affect resistance to innate immunity is critical when examining interactions with the host immune system.

Materials and Methods

Strains and Media.

A. actinomycetemcomitans strains VT1169 (44), Y4 (45), S. strain Challis DL1.1 (ATCC 49818), E. coli DH5α, and E. coli SM10 were used in this study. A. actinomycetemcomitans strains were grown in brain heart infusion medium, tryptic soy broth + 0.5% yeast extract (TSBYE), or chemically defined medium (CDM) with 20 mM glucose (15). Culture conditions were at 37 °C in a 10% (vol/vol) CO₂ atmosphere with shaking at 165 rpm unless otherwise indicated. E. coli strains were grown on LB at 37 °C. Where applicable, antibiotics were used at the following concentrations: chloramphenicol at 2 μg/mL for selection and maintenance in A. actinomycetemcomitans and at 20 μg/mL for selection and maintenance in E. coli. In both A. actinomycetemcomitans and E. coli, spectinomycin was used at 50 μg/mL for selection and at 10 μg/mL for maintenance, and streptomycin was used at 50 μg/mL for selection and at 20 μg/mL for maintenance.

DNA and Plasmid Manipulations.

DNA and plasmid isolations were performed using standard methods (46).

GeneChip and RT-PCR Analysis.

For flow cell biofilm experiments, cells were grown in 20% (vol/vol) TSBYE medium in a once-flow-through biofilm flow cell as described (47). Biofilms were allowed to mature for 18 h, and TSBYE, with or without 1 mM H₂O₂, was then added for 30 min. To harvest biofilm cells, the coverslip was removed from the flow cell with a razorblade and vortexed for 1 min in 20 mL of RNALater to remove attached cells. For colony biofilms, 10⁵ cells were spotted onto a UV-sterilized 0.2-μm polycarbonate membrane on 10 mL of solid CDM containing 1.5% (vol/vol) agarose in a 100-mm Petri dish. Cells were then grown for 32 h at 37 °C in a 10% (vol/vol) CO₂ atmosphere. The membranes were transferred to identical Petri plates and incubated for 2 h before being transferred to another Petri plate containing CDM with or without the addition of 1 mM H₂O₂. After 20 min, the membranes were transferred into 20 mL of RNALater. Cells were harvested from the membranes by gentle vortexing in the RNALater solution for ≈2 min until no cells visibly remained on the membrane. RNA isolation, preparation of labeled cDNA, and processing of the A. actinomycetemcomitans GeneChip microarrays were performed as described previously (48). Data analysis was performed using GeneChip Operating Software version 1.4 (Affymetrix). RT-PCR was performed as described (47) with the following changes: 100 ng of RNA was used for cDNA synthesis, 1 ng of cDNA was used as a template in the katA and apiA PCR reactions, and 5 ng of cDNA was used as a template in the clpX PCR. RT-PCR primers are included in SI Text. Planktonic A. actinomycetemcomitans was grown to the midexponential phase (OD₆₀₀ = 0.4) and mixed 1:1 with RNALater before RNA purification.

Primer Extension.

Primer extension was performed as previously described (49). Primers used were apiA-PE (5′-tctttagcccaatgcattgacaga-3′) and katA-PE (5′-catggtgttgtcattatcca-3′). The sizes of primer extension products were determined at the University of Oklahoma Health Science Center sequencing core facility.

Luminescence Reporter Assays.

A total of 10⁷ A. actinomycetemcomitans carrying either the apiA-luxCDABE or katA-luxCDABE reporter (reporter construction described in SI Text) was spread evenly over the surface of a TSBYE agar plate and grown overnight at 37 °C. Two 0.6-cm paper disks containing 10⁷ S. gordonii were added to each A. actinomycetemcomitans–coated TSBYE agar plate. One disk received 10,000 U of bovine catalase (Sigma), and the other disk received 10,000 U of heat-killed catalase. The plates were incubated for an additional 4 h at 37 °C before imaging each for 15 min with a Syngene G:Box (Syngene) imaging system.

Construction and Complementation of an oxyR⁻ Mutant in A. actinomycetemcomitans.

Construction and complementation of the oxyR⁻ mutant were performed as previously described (15) and are detailed in SI Text.

Serum Sensitivity Factor H Binding.

A. actinomycetemcomitans serum sensitivity was determined as previously described (26) with minor modifications. Details of this method can be found in SI Text. A. actinomycetemcomitans and S. gordonii cocultures were grown as described for the serum sensitivity assay described previously. After 30 min of coculture, 100 μL of human serum was added to 900 μL of coculture and incubated at 37 °C for 30 min. Cells were collected by centrifugation at 5,000 × g for 10 min, and cell pellets were resuspended in 100% (vol/vol) ice-cold methanol and fixed at −20 °C for 2 h. After fixing, cells were stained as described (50) using 1:1,000 mouse anti-human factor H antibody (Santa Cruz Biotechnology) and 1:1,000 Alexa-488 goat anti-mouse (Invitrogen) secondary antibody. Images were captured at a magnification ×1,000 using a Nikon 50i microscope, 100 × 1.4NA PLAN APO lens, Nikon DS-2MBW digital camera, and Nikon NIS-Elements D 3.0 software.